See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/221965896

Direct Observation of Interband Spin-Orbit

Coupling in a Two-Dimensional Electron System

Article in Physical Review Letters March 2012

DOI: 10.1103/PhysRevLett.108.196801 Source: arXiv

CITATIONS READS

22 50

5 authors, including:

Hendrik Bentmann Samir Abdelouahed

University of Wuerzburg The University of York
28 PUBLICATIONS 226 CITATIONS 10 PUBLICATIONS 152 CITATIONS

SEE PROFILE SEE PROFILE

Mattia Mulazzi
Humboldt-Universitt zu Berlin
31 PUBLICATIONS 239 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Hendrik Bentmann
letting you access and read them immediately. Retrieved on: 31 October 2016
 Direct Observation of Interband Spin-Orbit Coupling in a Two-Dimensional Electron System

Hendrik Bentmann,1, 2 Samir Abdelouahed,3 Mattia Mulazzi,1, 2 Jurgen Henk,4 and Friedrich Reinert1, 2
1
Experimentelle Physik VII and Rontgen Research Center for Complex Materials (RCCM),
Universitat Wurzburg Am Hubland, D-97074 Wurzburg, Germany
2
Karlsruhe Institute of Technology KIT, Gemeinschaftslabor fur Nanoanalytik, D-76021 Karlsruhe, Germany
3
Texas A&M University at Qatar, P.O. Box 23874, Doha, Qatar
4
Institut fur Physik - Theoretische Physik, Martin-Luther-Universitat Halle-Wittenberg, D-06099 Halle (Saale), Germany
(Dated: March 29, 2012)
arXiv:1203.6190v1 [cond-mat.mes-hall] 28 Mar 2012

We report the direct observation of interband spin-orbit (SO) coupling in a two-dimensional (2D) surface
electron system, in addition to the anticipated Rashba spin splitting. Using angle-resolved photoemission exper-
iments and first-principles calculations on Bi/Ag/Au heterostructures we show that the effect strongly modifies
the dispersion as well as the orbital and spin character of the 2D electronic states, thus giving rise to considerable
deviations from the Rashba model. The strength of the interband SO coupling is tuned by the thickness of the
thin film structures.

The spin-orbit interaction plays a fundamental role in the
rapidly developing field of spintronics as it allows for the
electrostatic manipulation of the spin degrees of freedom in
the conduction channels of nanoscale heterostructures [1, 2].
Such applications are often based on the Rashba effect in a
two-dimensional electron gas (2DEG) that prescribes a lift-
ing of the spin degeneracy in the presence of structural in-
version asymmetry and strong spin-orbit (SO) coupling [3].
Other contributions to the spin splitting arise from the Dres-
selhaus effect for constituent bulk crystal structures without
a center of inversion [4]. Moreover, new effects due to the
SO interaction have been discovered recently that give rise to
topologically protected, spin-polarized states on the surfaces
of a number of heavy-element semiconductors, referred to as
topological insulators [5, 6].
In the phenomenological treatment of the mentioned ef-
fects the spin-polarized electronic states are usually assumed
as pure spin states. In a real system, on the other hand, the SO
interaction couples spin and orbital angular momentum which
will result in a mixing of orthogonal spinors in the single-
particle eigenstates [7]. Recent ab-initio calculations suggest
that this can result in considerable reductions of the spin po-
larization of spin-split two-dimensional electronic states in the
presence of strong SO interaction [8]. Even more profound
effects of spin-mixing are known from the three-dimensional
(3D), exchange-split band structures of ferromagnets where
the SO coupling allows for hybridizations between spin-up
and spin-down bands. This interband SO coupling lies at the
origin of several magnetic phenomena, e.g. magneto crys-
talline anisotropy [9] or ultrafast demagnetization [10]. Note
that spin-mixing is expected to gain increasingly in impor-
tance for materials with high atomic number and thus en-
hanced SO interaction. It is therefore of fundamental impor-
tance to explore whether interband SO coupling effects due
to spin-mixing are present in heavy-element 2DEGs and how
they modify the spin-split electronic structure. Indeed, recent
theoretical reports predict interband SO coupling phenomena
in 2DEGs formed in zinc blende quantum wells [11, 12] and
on high-Z metal surfaces [13] as a result of higher-order per-
turbation theory corrections in the SO interaction. Yet, to best
of our knowledge, these effects have not been addressed in
experiment so far.
In this Letter we report the direct observation of inter-
band SO coupling in a 2DEG with large Rashba splitting
that is formed in a BiAg2 surface alloy grown on Ag quan-
tum films supported by a Au(111) substrate [Fig. 1(b)]. Us-
ing angle-resolved photoelectron spectroscopy (ARPES) with
high energy-resolution we find avoided crossings in the spin-
split electronic structure that provide evidence for the hy-
bridization of states with opposite spin due to SO coupling.
These findings are in line with relativistic first-principles com-
putations and model calculations. Further, we demonstrate
that the strength of the interband SO effect varies upon chang-
ing the thickness of the Ag film whereas the Rashba cou-
pling remains unmodified. The new interband coupling hence
emerges to be tunable, independently from other SO effects,
by nanostructural design capabilities as shown here for the
layered Ag/Au heterostructure. Our investigations prove that
the SO interaction can considerably influence the electronic
states in 2D systems, in addition to established effects such
as Rashba splitting. We hence expect the present findings
to be highly relevant for spintronic applications based on
strongly SO coupled compounds including the recently re-
ported heavy-element semiconductors with large Rashba split-
ting [14, 15] and topological insulators [16].
The BiAg2 system and related isostructural alloys have
been shown previously to feature an unusually large Rashba
effect in their electronic structure resulting from the strong SO
interaction of the Bi atoms [17, 18, 19, 20]. The band structure
of the surface alloys consists of two parabolic, Rashba-split
states E1,2 with negative effective mass [Fig. 1(a)], whose
spin-polarization has been verified by spin-resolved photoe-
mission experiments [21, 22, 23].
The ARPES data were collected by a SCIENTA R4000
electron spectrometer employing a monochromatized He dis-
charge lamp operating at an excitation energy of 21.22 eV
(He I ). The energy and angular resolution of the setup are
E = 3 meV and = 0.3 , respectively. We performed all
measurements at base pressures lower than 2 1010 mbar
and at temperatures of 20 K. The preparation of the Au(111)
 2

(a) (b) spin-splitting around the point, and a backfolded Au bulk

E2 Rashba
Interband
BiAg2 band u. The measured surface band structure thus complies
with the scheme in Fig. 1(a) with a position of the Fermi en-
Energy EF
Ag film ergy as indicated. Deviations from the pure Rashba scenario
E1

[full line in Fig. 1(a)] are observed in the experimental data
Au(111) at the expected crossing points between the two branches E1+
(c)
Wave vector
and E2 . Near these points the two states hybridize which
+
u E2 E2
leads to a gap opening and pronounced, kink-like changes in
0.0 the dispersion [highlighted by the circle in (c)]. Note that
+
E1 these observations cannot be explained within the Rashba
E-EF [eV]

-0.2
-0.4
E1
-0.6
-0.4 -0.2 0.0
Wave vector kx [1/]
FIG. 1: (Color online) (a) Band structure of a two-band system with
Rashba interaction (schematic). Two different cases with (dashed
line) and without (full line) an additional interband spin-orbit cou-
pling are shown. (b) Sketch of the studied multilayer sample: The
BiAg2 layer is grown on a Ag film of variable thickness supported
by the Au(111) substrate. (c) Experimental band structure for BiAg2
on 4 ML Ag/Au(111) along the K direction of the surface Brillouin
zone. The dashed line indicates the edge of the projected bulk band
gap of Au(111).
substrate by cycles of Ar sputtering and annealing resulted in
a clean and well-ordered surface as verified by the spectral
linewidth of the Au(111) surface state [24]. Ag was evap-
orated on the cooled substrate at 150 K. Subsequent mild
annealing resulted in homogeneous films as confirmed by the
characteristic shift in binding energy of the surface state for
low Ag coverages [25]. The BiAg2 surface alloy was obtained
after evaporation of 1/3 ML Bi on the substrate film at ele-
vated temperatures of 400 K. We observed no change in the
model which assigns pure opposite spin states to these two
bands. A hybridization between them should be prohibited
in this case. The findings are therefore distinctively different
from the previously found hybridization phenomena between
Rashba-split states and spin-degenerate quantum well states
0.2 0.4
[22, 29, 30]. Our experimental results indicate an additional
interband SO effect that couples orbitals with opposite spin
and thereby induces the hybridization.
To gain a first understanding of the observed effect we
adopt an effective 2D model Hamiltonian taking into account
Rashba and interband contributions of the SO coupling but
neglecting the orbital part of the wave functions for the mo-
ment, similar as it is done for the Rashba model in single band
systems. The interband SO coupling term introduces a fi-
nite hybridization = hE1+ |Hso |E2 i between the two purely
Rashba-split states. For the modified
q eigenvalues S we then
have S = 21 (E1+ + E2 ) 14 (E1+ E2 )2 + 2 , where
we omit the k-dependence for clarity. Close to the crossing
point of E1+ and E2 the SO coupling mixes the states |E1+ i
and
p |E2 i resulting in new eigenstates |S+ i = ak |E1+ i +
1 ak |E2 i and |S i = 1 ak |E1+ i ak |E2 i. The
p
2 2
(S E + )2
coefficient ak can be expressed by |ak |2 = (1+ + 2 1 )1 .
Our experimental data allow us not only to determine the
(a) (b)
Photoemission intensity [a.u.]

binding energy of the Shockley state for Ag/Au(111) at this 0.0

1.0
Imax
temperature providing evidence for no or neglegible Ag/Au S+ 0.8
intermixing at the interface [26]. Low energy
electron diffrac-
E-EF [eV]

-0.1 I
tion measurements confirmed the ( 3x 3) reconstruction of 0.6
I+
the surface alloy. 0.4
-0.2
-
First-principles calculations were carried out using the S
0.2 Imin
relativistic full-potential linearized augmented plane wave
-0.3
(FLAPW) method as implemented in the Fleur code [27]. Ex- -0.20 -0.16 -0.12 -0.08 0.18 0.16 0.14 0.12 0.10
change and correlation were treated within the generalized Wave vector kx [1/] Wave vector kx [1/]
gradient approximazion (GGA) [28]. We used a plane-wave
cutoff of 8.6 A 1 while the charge density and potential cut- FIG. 2: (Color online) Comparison of the experimental electronic
offs were 21.9 A 1 . A ten-layer slab (BiAg2 /4ML Ag/5 ML structure close to the SO-induced hybridization gap with the solution
Au) was used to simulate the surface electronic structure. The of a model Hamiltonian (see text). A closeup of the hybridization
gap taken from the dataset in Fig. 1(c) is shown in (a). The ARPES
opposite slab surface was saturated by hydrogen atoms to sup-
spectrum is overlaid by the modelled dispersions S (dark lines) for
press the formation of the Au(111) surface state. The vertical an interaction strength = 31 meV. The scale bar above the graph
relaxation of the Bi atoms was 0.95 A. indicates the k-interval of panel (b). In (b) the datapoints mark the
Fig. 1(c) shows the experimental electronic structure of a measured k-dependent intensity evolution I of the two branches
BiAg2 surface alloy grown on a 4 ML Ag film on Au(111). S . The experimental results are compared with the calculated in-
We identify two pairs of states E1,2
, both showing Rashba tensities (full lines) based on the model Hamiltonian.
 3

periment (0.01 A 1 ). The full lines in (b) correspond to the

expected intensity change according to the model Hamiltonian
1 using the same value for and the same dispersion relations
+
E2 as in (a). Again, we find a good agreement with the experi-
mental data. We hence conclude that the observed dispersion
-
E2 modification and change in state character can be consistently
E - EF (eV)

+ described by a Rashba-type free electron model including an

E1
0 additional interband SO coupling term. Comparing the en-
-
E1
-1-0.2
-0.1 -1 0.1 0.2
kx[ ]
FIG. 3: (Color online) Spin-resolved surface electronic struture
of BiAg2 on 4 ML Ag/Au(111) as obtained by a first-principles
calculation. The size of the markers scales with the degree of
spin-polarization of the electronic states and the color refers to the
spin-orientations up (red,light) and down (blue,dark) with the spin-
quantization axis along the y-direction. The hybridization gap
between the branches E1+ and E2 is indicated.
modified dispersions S but also to trace the k-dependence
of the coefficient |ak |2 [Fig. 2]. The latter is possible due
to considerably differing photoionization cross sections for
the two states |E1+ i (high cross section) and |E2 i (low cross
section). As a result, the branches S show drastic intensity
changes close to the hybridization gap where their state char-
ergy scales of the two SO effects in this system we find the
interband contribution (30 meV) to be about one order of
magnitude weaker than the Rashba contribution (300 meV).
Qualitative insight into the coupled orbital and spin parts of
the electronic states |E1+ i and |E2 i which goes beyond the
discussed free electron model can be attained by inspecting
the spatial symmetry and the contributing atomic orbitals in
the system. Group-theoretical considerations show that both
states belong to the same representation and that their wave
function along kx can be written as |i = |spz i + |px
i + |py i, where the spinors are quantized with respect to
the ky axis [13]. It is important to note that the SO interac-
tion induces a mixing of opposite spinors in the wave func-
tion and therefore facilitates a hybridization between the two
bands which is indeed observed in our measurements. The hy-
bridization mechanism is therefore comparable to the effect of
SO coupling on the 3D spin-split band structures in ferromag-
2 ML
0.00 S+
E-EF [eV]

acter is strongly modified. We identify the k-dependent inten- -0.10

sity evolution I of the branches S with the relative contri-
butions of |E1+ i and |E2 i to |S i. These contributions are -0.20 S
directly given by ak and we have the simple correspondence -0.30
I+ = Imin + (Imax Imin )|ak |2 and accordingly for I .
Within this approximation we neglect photoemission matrix
0.00 4 ML
element effects. Note that ak and I depend on the hybridiza-
E-EF [eV]

tion strength . -0.10

A comparison of the experimental data with the results of
-0.20
the model Hamiltonian is shown in Fig. 2. In (a) we dis-
play the measured band structure and the model dispersions -0.30
close to the hybridization gap. The bands E1+ and E2 [red
(light) lines in Fig. 2(a)] are determined from the experimen- 0.00 16 ML
tal data in (E, k) regions sufficiently far away from the hy-
E-EF [eV]

bridization gap. For E2 we use a linear dispersion which -0.10

is an adequate approximation for the small (E, k) window
-0.20
which is of interest here. A close match between the experi-
mental bands and S is obtained for = 31 meV. Fig. 2(b) -0.30
shows the photoemission intensities I obtained from energy
and momentum distribution curves as a function of the wave -0.20 -0.15 -0.10 -0.05 0.00
vector kx . The data for each branch were normalized to the Wave vector kx [1/]
respective maximal value of the individual data set. For in-
creasing wave vector kx the intensity I+ is enhanced whereas FIG. 4: Modification of the SO-induced hybridization gap arising
I is reduced reflecting the change in state character of both from changes in the thickness of the supporting Ag film. The three
branches near the hybridization gap. We confirmed that this panels show second-derivative spectra of the experimental data for
behavior is not affected by the finite k-resolution of the ex- Ag layers of 2 ML, 4 ML and 16 ML.

netic systems [10, 31].
To further substantiate the experimental results on a quan-
titative theoretical level we examine the spin-polarized elec-
tronic structure by a first-principles calculation [Fig. 3]. In
accordance with experiment our calculation finds two Rashba-
split states and a hybridization gap between the two branches
E1+ and E2 . Similar to the measured results the strength of
interband coupling is smaller than that of the Rashba cou-
pling. Note that the spin-polarization of the branch E2 shows
a sign change near the hybridization kink clearly reflecting the
mixed and k-dependent spin character of the corresponding
wave function. We further find the degree of spin-polarization
to be reduced from 100 %, as predicted by the Rashba model,
to maximal values of 66 %.
It is interesting to explore the possibility to modify the
electronic structure of the BiAg2 alloy by choosing a differ-
ent thickness for the supporting Ag quantum well film. We
have recently shown that the spatial localization of quantum
well states in the system Ag/Au(111) changes considerably
as a function of layer thickness, especially in the low cover-
age regime up to 20 ML [32]. As a result the charge den-
sity close to the surface, where the BiAg2 alloy is located,
varies strongly depending on the precise Ag film thickness.
Hence, one may expect concomitant influences of these vari-
ations on the electronic states in the surface alloy. Indeed, we
find such changes as is inferred from Fig. 4 which shows the
electronic structure of BiAg2 close to the hybridization gap for
three different Ag layers. Whereas the general features in the
electronic structure, such as binding energy and Rashba split-
ting, are very similar for all three films we find considerable
changes in the size of the hybridization gap and therefore
in the interband coupling strength. The largest gap is found
for 2 ML ( = 42 meV) whereas at higher thicknesses it is
reduced ( = 14 meV for the case of 16 ML). The strength of
the interband SO coupling is therefore determined by details
of the geometric and electronic substrate properties. This pro-
vides possibilities to tailor this new SO effect by a controlled
variation of the corresponding parameters in custom designed
nanostructures, as exemplified here by the electronically tun-
able quantum well system Ag/Au.
We have established the presence of interband SO coupling
in 2DEGs with Rashba spin splitting by photoemission spec-
troscopy experiments on multilayered Bi/Ag/Au heterostruc-
tures. The coupling induces hybridization between bands of
different spin polarizations and thereby causes considerable
deviations from the Rashba model in the dispersion of the
2D electronic states as well as in their spin and orbital mo-
mentum character. Such effects will certainly influence the
transport properties in 2DEGs, especially in compounds with
very strong SO interactions, and may therefore be exploited
to improve the performance of new spintronic functionalities
[33, 34].
HB acknowledges helpful discussions with Frank Forster
and Kazuyuki Sakamoto. This work was supported by the
Bundesministerium fur Bildung und Forschung (Grants No.
05K10WW1/2 and No. 05KS1WMB/1) and the Deutsche
4
Forschunsgsgemeinschaft within the Forschergruppe 1162
and the Sonderforschungsbereich 762.
c, J. Fabian, and S. Das Sarma, Rev. Mod. Phys. 76, 323
[1] I. Zuti
(2004).
[2] H. C. Koo, J. H. Kwon, J. Eom, J. Chang, S. H. Han, and
M. Johnson, Science 325, 1515 (2009).
[3] Y. A. Bychkov and E. I. Rashba, JETP Lett. 39, 78 (1984).
[4] G. Dresselhaus, Phys. Rev. 100, 580 (1955).
[5] L. Fu and C. L. Kane, Phys. Rev. B 76, 045302 (2007).
[6] D. Hsieh, Y. Xia, L. Wray, D. Qian, A. Pal, J. H. Dil, J. Oster-
walder, F. Meier, G. Bihlmayer, C. L. Kane, et al., Science 323,
919 (2009).
[7] J. Henk, T. Scheunemann, S. V. Halilov, and R. Feder, Journal
of Physics: Condensed Matter 8, 47 (1996).
[8] O. V. Yazyev, J. E. Moore, and S. G. Louie, Phys. Rev. Lett.
105, 266806 (2010).
[9] P. Bruno, Phys. Rev. B 39, 865 (1989).
[10] M. Pickel, A. B. Schmidt, F. Giesen, J. Braun, J. Minar,
H. Ebert, M. Donath, and M. Weinelt, Phys. Rev. Lett. 101,
066402 (2008).
[11] E. Bernardes, J. Schliemann, M. Lee, J. C. Egues, and D. Loss,
Phys. Rev. Lett. 99, 076603 (2007).
[12] R. S. Calsaverini, E. Bernardes, J. C. Egues, and D. Loss, Phys.
Rev. B 78, 155313 (2008).
[13] H. Mirhosseini, J. Henk, A. Ernst, S. Ostanin, C.-T. Chiang,
P. Yu, A. Winkelmann, and J. Kirschner, Phys. Rev. B 79,
245428 (2009).
[14] K. Ishizaka, M. S. Bahramy, H. Murakawa, M. Sakano,
T. Shimojima, T. Sonobe, K. Koizumi, S. Shin, H. Miyahara,
A. Kimura, et al., Nat Mater 10, 521 (2011).
[15] P. D. C. King, R. C. Hatch, M. Bianchi, R. Ovsyannikov,
C. Lupulescu, G. Landolt, B. Slomski, J. H. Dil, D. Guan, J. L.
Mi, et al., Phys. Rev. Lett. 107, 096802 (2011).
[16] Y. Xia, D. Qian, D. Hsieh, L. Wray, A. Pal, H. Lin, A. Bansil,
D. Grauer, Y. S. Hor, R. J. Cava, et al., Nat. Phys. 5, 398 (2009).
[17] D. Pacile, C. R. Ast, M. Papagno, C. Da Silva, L. Moreschini,
M. Falub, A. P. Seitsonen, and M. Grioni, Phys. Rev. B 73,
245429 (2006).
[18] C. R. Ast, J. Henk, A. Ernst, L. Moreschini, M. C. Falub,
D. Pacile, P. Bruno, K. Kern, and M. Grioni, Phys. Rev. Lett.
98, 186807 (2007).
[19] L. Moreschini, A. Bendounan, H. Bentmann, M. Assig,
K. Kern, F. Reinert, J. Henk, C. R. Ast, and M. Grioni,
Phys. Rev. B 80, 035438 (2009).
[20] H. Bentmann, F. Forster, G. Bihlmayer, E. V. Chulkov,
L. Moreschini, M. Grioni, and F. Reinert, EPL 87, 37003
(2009).
[21] F. Meier, H. Dil, J. Lobo-Checa, L. Patthey, and J. Osterwalder,
Phys. Rev. B 77, 165431 (2008).
[22] K. He, Y. Takeichi, M. Ogawa, T. Okuda, P. Moras, D. Topwal,
A. Harasawa, T. Hirahara, C. Carbone, A. Kakizaki, et al., Phys.
Rev. Lett. 104, 156805 (2010).
[23] H. Bentmann, T. Kuzumaki, G. Bihlmayer, S. Blugel, E. V.
Chulkov, F. Reinert, and K. Sakamoto, Phys. Rev. B 84, 115426
(2011).
[24] G. Nicolay, F. Reinert, S. Hufner, and P. Blaha, Phys. Rev. B
65, 033407 (2001).
[25] D. Popovic, F. Reinert, S. Hufner, V. G. Grigoryan, M. Spring-
borg, H. Cercellier, Y. Fagot-Revurat, B. Kierren, and D. Mal-

terre, Phys. Rev. B 72, 045419 (2005).
[26] H. Cercellier, C. Didiot, Y. Fagot-Revurat, B. Kierren,
L. Moreau, D. Malterre, and F. Reinert, Phys. Rev. B 73,
195413 (2006).
[27] For program description see http://www.flapw.de.
[28] J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77,
3865 (1996).
[29] K. He, T. Hirahara, T. Okuda, S. Hasegawa, A. Kakizaki, and
I. Matsuda, Phys. Rev. Lett. 101, 107604 (2008).
[30] E. Frantzeskakis, S. Pons, H. Mirhosseini, J. Henk, C. R. Ast,
and M. Grioni, Phys. Rev. Lett. 101, 196805 (2008).
5
[31] A. Rampe, G. Guntherodt, D. Hartmann, J. Henk, T. Scheune-
mann, and R. Feder, Phys. Rev. B 57, 14370 (1998).
[32] F. Forster, E. Gergert, A. Nuber, H. Bentmann, L. Huang, X. G.
Gong, Z. Zhang, and F. Reinert, Phys. Rev. B 84, 075412
(2011).
[33] J. C. Egues, G. Burkard, and D. Loss, Applied Physics Letters
82, 2658 (2003).
[34] J. C. Egues, G. Burkard, and D. Loss, Phys. Rev. Lett. 89,
176401 (2002).